To Bet or Not to Bet
The follower of this blog will remember that I covered the 2008 elections' betting site on Rasmussen Markets (not real money, though), in which one could take sides on a market for various outcomes--primary election winners, the Presidential race, how many seats the Democrats would win in Congress, and individual races as well. There is some experimental data that things like this can be good oracles of probable outcomes for things like this, or for something like the future performance of the stock--the so-called Delphi effect, but this requires large numbers of people and an active market.
After the 2008 election, though, Rasmussen shut the site down. It was never explained, but I guess that the site did not do what Rasmussen wanted, which was provide some independent confirmation of what they were seeing in the polls, and that failure might have been ascribed to the fact that real money was not at stake. Thus, political operatives or others with ideological axes to grind might have tried to manipulate the results.
There still is Intrade, though, and I have revived my account there so I could monitor the action. Intrade is based somewhere offshore of the British Isles, and it is technically illegal for Americans to wager there (prohibitions on online betting)--if you have an offshore account you can fund it, though they won't accept American credit cards. It is not illegal to have an account there, though.
Intrade makes markets on all kinds of things--sports, the overall performance of the stock market, movie box office or Oscars--both short-term and long-term. Elections, though, are the basis of a very extensive variety of posted markets there--this was true, for example, of the recent British and Brazilian elections, too. Some of those markets are very poorly populated, as measured by number of contracts, or number of bidders out there, while others have thousands or tens of thousands of contracts booked. Most--though not all--of the ones booked on the American elections will be resolved tonight, though there can be dramatic moves in prices of new trades, sometimes several dramatic moves, in the course of an evening. It's also interesting to follow how these "markets" have moved over time; often in synch with polling results, naturally, but sometimes much more dramatically, and sometimes surprisingly.
My feeling about the practicalities of investment on these bets is that one shouldn't/wouldn't take a position unless one has a strong feeling to one side or the other. Of course, there is the temptation to play in the big contests for entertainment purposes.
Getting Down to Cases
Just as I saw with Rasmussen, or one might see with a poll of Wall Street traders, I feel that the Intrade-playing population is a bit biased toward the Republicans. So, for example, while 538.com rates the chances of the Democrats holding onto a majority at 16%, you can get that contract at 4.8 out of 100 (Republicans winning is at 97.0, and "neither", which by the rules would mean that neither side had 218 official winners by Dec. 31, was last traded at 0.5). Now, I would not recommend buying that Democratic-win contract, even if it's underpriced, unless one is willing to stay with the screen and sell it at the first positive news for Democrats during the evening. I learned from my Rasmussen Markets experience that there can be such movements, and that even if things might look good for awhile, you don't want to be hanging onto the losing end of a contract at any price.
The House bet that I do like is on the number of seats the Republicans pick up: Silver (and a consensus of various experts) are talking about 55 or so as the median of their probability distributions, but for that bet (that Republicans pick up 55 or more seats) you'd have to pay 71.0 (last trade, and also current lowest "ask" price), so that might be one to bet against. That "71" is a substitute for the probability of the outcome, the odds at which buyers and sellers balance out. Even better might be to bet against the Republicans gaining 50 or more seats: that last traded at 88, though the price seems to be moving down (you could sell a contract for 82.0 right now, while the low "ask" price now is at 87). The idea would be to bet against the Republicans' doing that, but to be prepared to close out the position, buying an equal amount in the other direction at a lower price. Timing would be critical, and at some point you'd have to decide whether the trend would hold until the end, making you a winner, or would you need to close it out and (hopefully) take a profit, and it may take several nervous days for the contract to be finally decided, if that turns out to be close to the final number (if it's not, trading interest will dry up completely even if the contract is not definitively judged to be either a winner or loser).
When it comes to the Senate, the way Intrade has posted the rules has made for some interesting plays. They are not counting independents Lieberman and Sanders toward the Democrats' total, so the bet is whether the Democrats have 50 votes (because the V.P. would break a tie) without them. As for the bet on the Republican side, you could choose whether to bet they have actual control (51 seats, again without any independents--Murkowski in Alaska, presumably, wouldn't count) or whether they have 50 or more seats. Then, there is the "neither" bet, which would finish "in the money" if the Democrats have 50 or 51 seats (counting Lieberman/Sanders, or 48 or 49 without them). This bet is very popular, with very narrow spreads between bid and ask: most recent trade quotes (as of 9 a.m. MDT) are 48 for Democratic control (should be interpreted in the conventional sense as 52 or more Democratic caucus votes), 12.5 for Republicans, and 42.8 for "Neither"!
The probability, as described above, for a "neither" outcome would have to be rated pretty high--essentially if every race goes as current polls indicate, the Democrats would have 51 and it would be a winner--but the variability on outcomes at this point would suggest this is going to be a very volatile bet through the night. This makes it a good trade, but not necessarily one to "buy and hold".
When it comes to individual races, I find the prices fairly realistic when there is sufficient market volume. Here are some current quotes on individual Senate candidates' chances of winning: Bennet of Colorado, 30.0; Boxer of California, 90.0; Marco Rubio of Florida, 93.0; Kirk in Illinois, 78.9; and Sestak of Pennsylvania, 14.0 (but dropping). Two with somewhat unusual quotes at present are Nevada, where Senator Reid's last quote is at 34, and his opponent Sharron Angle is at 75 (suggesting some strong support for both sides, and a great opportunity for arbitrage), and the difficult race to handicap in Alaska (because of the strong write-in candidacy of the incumbent Lisa Murkowski, and the possibility that support for Republican candidate Joe Miller may be dropping): current quotes are 77.5 for the Republican Miller, 8.0 for the Democratic underdog MacAdams, and 26.5 for Neither (a/k/a Murkowski); however, there are very wide spreads between bid and ask on some of those trades, suggesting there will be a lot of volatility--and that may be true for days or weeks, as the write-in votes will need to be individually counted, so if the race is close, these may continue to fluctuate. Or not.
The key races in the House are generally not so heavily traded, and, somewhat surprisingly, that's true of the governors' races, too, so probably they are not worth commenting upon.
In a longer-term sense, you can bet on whether Sarah Palin will get the Republican nomination in 2012 (currently around 19) or Barack Obama will win re-election (61). These quotes both seem pretty reasonable to me.